From 66acb76bb0492b263215ca9b4d927a7be39ace02 Mon Sep 17 00:00:00 2001 From: Anton Khirnov Date: Tue, 16 Jun 2015 18:22:11 +0200 Subject: lavc: add Intel libmfx-based HEVC encoder --- libavcodec/qsvenc_hevc.c | 272 +++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 272 insertions(+) create mode 100644 libavcodec/qsvenc_hevc.c (limited to 'libavcodec/qsvenc_hevc.c') diff --git a/libavcodec/qsvenc_hevc.c b/libavcodec/qsvenc_hevc.c new file mode 100644 index 0000000000..03007f1814 --- /dev/null +++ b/libavcodec/qsvenc_hevc.c @@ -0,0 +1,272 @@ +/* + * Intel MediaSDK QSV based HEVC encoder + * + * This file is part of Libav. + * + * Libav is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * Libav is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with Libav; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + + +#include +#include + +#include + +#include "libavutil/common.h" +#include "libavutil/opt.h" + +#include "avcodec.h" +#include "bytestream.h" +#include "get_bits.h" +#include "hevc.h" +#include "internal.h" +#include "qsv.h" +#include "qsv_internal.h" +#include "qsvenc.h" + +enum { + LOAD_PLUGIN_NONE, + LOAD_PLUGIN_HEVC_SW, + LOAD_PLUGIN_HEVC_HW, +} LoadPlugin; + +typedef struct QSVHEVCEncContext { + AVClass *class; + QSVEncContext qsv; + int load_plugin; +} QSVHEVCEncContext; + +static int generate_fake_vps(QSVEncContext *q, AVCodecContext *avctx) +{ + GetByteContext gbc; + PutByteContext pbc; + + GetBitContext gb; + HEVCNAL sps_nal = { NULL }; + HEVCSPS sps = { 0 }; + HEVCVPS vps = { 0 }; + uint8_t vps_buf[128], vps_rbsp_buf[128]; + uint8_t *new_extradata; + unsigned int sps_id; + int ret, i, type, vps_size; + + if (!avctx->extradata_size) { + av_log(avctx, AV_LOG_ERROR, "No extradata returned from libmfx\n"); + return AVERROR_UNKNOWN; + } + + /* parse the SPS */ + ret = ff_hevc_extract_rbsp(avctx->extradata + 4, avctx->extradata_size - 4, &sps_nal); + if (ret < 0) { + av_log(avctx, AV_LOG_ERROR, "Error unescaping the SPS buffer\n"); + return ret; + } + + ret = init_get_bits8(&gb, sps_nal.data, sps_nal.size); + if (ret < 0) { + av_freep(&sps_nal.rbsp_buffer); + return ret; + } + + get_bits(&gb, 1); + type = get_bits(&gb, 6); + if (type != NAL_SPS) { + av_log(avctx, AV_LOG_ERROR, "Unexpected NAL type in the extradata: %d\n", + type); + av_freep(&sps_nal.rbsp_buffer); + return AVERROR_INVALIDDATA; + } + get_bits(&gb, 9); + + ret = ff_hevc_parse_sps(&sps, &gb, &sps_id, 0, NULL, avctx); + av_freep(&sps_nal.rbsp_buffer); + if (ret < 0) { + av_log(avctx, AV_LOG_ERROR, "Error parsing the SPS\n"); + return ret; + } + + /* generate the VPS */ + vps.vps_max_layers = 1; + vps.vps_max_sub_layers = sps.max_sub_layers; + memcpy(&vps.ptl, &sps.ptl, sizeof(vps.ptl)); + vps.vps_sub_layer_ordering_info_present_flag = 1; + for (i = 0; i < MAX_SUB_LAYERS; i++) { + vps.vps_max_dec_pic_buffering[i] = sps.temporal_layer[i].max_dec_pic_buffering; + vps.vps_num_reorder_pics[i] = sps.temporal_layer[i].num_reorder_pics; + vps.vps_max_latency_increase[i] = sps.temporal_layer[i].max_latency_increase; + } + + vps.vps_num_layer_sets = 1; + vps.vps_timing_info_present_flag = sps.vui.vui_timing_info_present_flag; + vps.vps_num_units_in_tick = sps.vui.vui_num_units_in_tick; + vps.vps_time_scale = sps.vui.vui_time_scale; + vps.vps_poc_proportional_to_timing_flag = sps.vui.vui_poc_proportional_to_timing_flag; + vps.vps_num_ticks_poc_diff_one = sps.vui.vui_num_ticks_poc_diff_one_minus1 + 1; + + /* generate the encoded RBSP form of the VPS */ + ret = ff_hevc_encode_nal_vps(&vps, sps.vps_id, vps_rbsp_buf, sizeof(vps_rbsp_buf)); + if (ret < 0) { + av_log(avctx, AV_LOG_ERROR, "Error writing the VPS\n"); + return ret; + } + + /* escape and add the startcode */ + bytestream2_init(&gbc, vps_rbsp_buf, ret); + bytestream2_init_writer(&pbc, vps_buf, sizeof(vps_buf)); + + bytestream2_put_be32(&pbc, 1); // startcode + bytestream2_put_byte(&pbc, NAL_VPS << 1); // NAL + bytestream2_put_byte(&pbc, 1); // header + + while (bytestream2_get_bytes_left(&gbc)) { + uint32_t b = bytestream2_peek_be24(&gbc); + if (b <= 3) { + bytestream2_put_be24(&pbc, 3); + bytestream2_skip(&gbc, 2); + } else + bytestream2_put_byte(&pbc, bytestream2_get_byte(&gbc)); + } + + vps_size = bytestream2_tell_p(&pbc); + new_extradata = av_mallocz(vps_size + avctx->extradata_size + FF_INPUT_BUFFER_PADDING_SIZE); + if (!new_extradata) + return AVERROR(ENOMEM); + memcpy(new_extradata, vps_buf, vps_size); + memcpy(new_extradata + vps_size, avctx->extradata, avctx->extradata_size); + + av_freep(&avctx->extradata); + avctx->extradata = new_extradata; + avctx->extradata_size += vps_size; + + return 0; +} + +static av_cold int qsv_enc_init(AVCodecContext *avctx) +{ + QSVHEVCEncContext *q = avctx->priv_data; + int ret; + + if (q->load_plugin != LOAD_PLUGIN_NONE) { + static const char *uid_hevcenc_sw = "2fca99749fdb49aeb121a5b63ef568f7"; + static const char *uid_hevcenc_hw = "e5400a06c74d41f5b12d430bbaa23d0b"; + + if (q->qsv.load_plugins[0]) { + av_log(avctx, AV_LOG_WARNING, + "load_plugins is not empty, but load_plugin is not set to 'none'." + "The load_plugin value will be ignored.\n"); + } else { + av_freep(&q->qsv.load_plugins); + + if (q->load_plugin == LOAD_PLUGIN_HEVC_SW) + q->qsv.load_plugins = av_strdup(uid_hevcenc_sw); + else + q->qsv.load_plugins = av_strdup(uid_hevcenc_hw); + + if (!q->qsv.load_plugins) + return AVERROR(ENOMEM); + } + } + + ret = ff_qsv_enc_init(avctx, &q->qsv); + if (ret < 0) + return ret; + + ret = generate_fake_vps(&q->qsv, avctx); + if (ret < 0) { + ff_qsv_enc_close(avctx, &q->qsv); + return ret; + } + + return 0; +} + +static int qsv_enc_frame(AVCodecContext *avctx, AVPacket *pkt, + const AVFrame *frame, int *got_packet) +{ + QSVHEVCEncContext *q = avctx->priv_data; + + return ff_qsv_encode(avctx, &q->qsv, pkt, frame, got_packet); +} + +static av_cold int qsv_enc_close(AVCodecContext *avctx) +{ + QSVHEVCEncContext *q = avctx->priv_data; + + return ff_qsv_enc_close(avctx, &q->qsv); +} + +#define OFFSET(x) offsetof(QSVHEVCEncContext, x) +#define VE AV_OPT_FLAG_VIDEO_PARAM | AV_OPT_FLAG_ENCODING_PARAM +static const AVOption options[] = { + { "async_depth", "Maximum processing parallelism", OFFSET(qsv.async_depth), AV_OPT_TYPE_INT, { .i64 = ASYNC_DEPTH_DEFAULT }, 0, INT_MAX, VE }, + { "avbr_accuracy", "Accuracy of the AVBR ratecontrol", OFFSET(qsv.avbr_accuracy), AV_OPT_TYPE_INT, { .i64 = 0 }, 0, INT_MAX, VE }, + { "avbr_convergence", "Convergence of the AVBR ratecontrol", OFFSET(qsv.avbr_convergence), AV_OPT_TYPE_INT, { .i64 = 0 }, 0, INT_MAX, VE }, + + { "load_plugin", "A user plugin to load in an internal session", OFFSET(load_plugin), AV_OPT_TYPE_INT, { .i64 = LOAD_PLUGIN_HEVC_SW }, LOAD_PLUGIN_NONE, LOAD_PLUGIN_HEVC_HW, VE, "load_plugin" }, + { "none", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = LOAD_PLUGIN_NONE }, 0, 0, VE, "load_plugin" }, + { "hevc_sw", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = LOAD_PLUGIN_HEVC_SW }, 0, 0, VE, "load_plugin" }, + { "hevc_hw", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = LOAD_PLUGIN_HEVC_HW }, 0, 0, VE, "load_plugin" }, + + { "load_plugins", "A :-separate list of hexadecimal plugin UIDs to load in an internal session", + OFFSET(qsv.load_plugins), AV_OPT_TYPE_STRING, { .str = "" }, 0, 0, VE }, + + { "profile", NULL, OFFSET(qsv.profile), AV_OPT_TYPE_INT, { .i64 = MFX_PROFILE_UNKNOWN }, 0, INT_MAX, VE, "profile" }, + { "unknown", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = MFX_PROFILE_UNKNOWN }, INT_MIN, INT_MAX, VE, "profile" }, + { "main", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = MFX_PROFILE_HEVC_MAIN }, INT_MIN, INT_MAX, VE, "profile" }, + { "main10", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = MFX_PROFILE_HEVC_MAIN10 }, INT_MIN, INT_MAX, VE, "profile" }, + { "mainsp", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = MFX_PROFILE_HEVC_MAINSP }, INT_MIN, INT_MAX, VE, "profile" }, + + { "preset", NULL, OFFSET(qsv.preset), AV_OPT_TYPE_INT, { .i64 = MFX_TARGETUSAGE_BALANCED }, 0, 7, VE, "preset" }, + { "fast", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = MFX_TARGETUSAGE_BEST_SPEED }, INT_MIN, INT_MAX, VE, "preset" }, + { "medium", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = MFX_TARGETUSAGE_BALANCED }, INT_MIN, INT_MAX, VE, "preset" }, + { "slow", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = MFX_TARGETUSAGE_BEST_QUALITY }, INT_MIN, INT_MAX, VE, "preset" }, + + { NULL }, +}; + +static const AVClass class = { + .class_name = "hevc_qsv encoder", + .item_name = av_default_item_name, + .option = options, + .version = LIBAVUTIL_VERSION_INT, +}; + +static const AVCodecDefault qsv_enc_defaults[] = { + { "b", "1M" }, + { "refs", "0" }, + // same as the x264 default + { "g", "250" }, + { "bf", "3" }, + + { "flags", "+cgop" }, + { NULL }, +}; + +AVCodec ff_hevc_qsv_encoder = { + .name = "hevc_qsv", + .long_name = NULL_IF_CONFIG_SMALL("HEVC (Intel Quick Sync Video acceleration)"), + .priv_data_size = sizeof(QSVHEVCEncContext), + .type = AVMEDIA_TYPE_VIDEO, + .id = AV_CODEC_ID_HEVC, + .init = qsv_enc_init, + .encode2 = qsv_enc_frame, + .close = qsv_enc_close, + .capabilities = CODEC_CAP_DELAY, + .pix_fmts = (const enum AVPixelFormat[]){ AV_PIX_FMT_NV12, + AV_PIX_FMT_QSV, + AV_PIX_FMT_NONE }, + .priv_class = &class, + .defaults = qsv_enc_defaults, +}; -- cgit v1.2.3